Background
There are elevated rates of suicidality in adults diagnosed with autism spectrum conditions (ASC) [
1‐
5]. However, suicidality in ASC is poorly understood, and there is a paucity of research exploring
why adults with ASC (henceforth, autistic adults) may be at increased risk [
6]. Although a number of studies have explored suicidality in autistic adults, no study has yet utilised a suicidality assessment tool with evidence of validity [
4,
7,
8]. Non-suicidal self-injury (NSSI) is a risk factor for suicide attempts in the general population [
9]. However, to our knowledge, only one study has ever explored NSSI in a small sample of autistic adults using a validated instrument but did not explore associations with suicidality [
10]. Clearly, it is crucial to better understand suicidality in autistic adults, and associated risk markers, using instruments with evidence of validity (albeit not yet in autistic adults). Given the paucity of literature in the area of suicide in ASC research, it is important to engage with the autistic community in the refinement of research priorities to speed up progress and benefit the end users of research [
11]. This is the aim of the current study.
Suicidal thoughts and behaviours are significantly increased in autistic adults compared to the general population and other clinical groups. In a large sample of 374 adults newly diagnosed with Asperger syndrome (AS; autism without language delay or intellectual disability), 66% had contemplated suicide, significantly higher than the general population (17%) and patients with psychosis (59%); 35% had planned or attempted suicide [
2], higher than previous estimates of attempted suicide in general and university populations (2.5–10%) [
12‐
14]. Only one study has ever explored whether autistic people are more at risk of dying by suicide than the general population; this population study in Sweden showed that autistic people were significantly more likely to die by suicide (0.31%) compared to the general population (0.04%) [
15].
Traits characteristic of autism are also significantly associated with suicidality in those with [
2], and without ASC diagnosis [
16‐
18]. ASC diagnosis has also recently been found to be an independent risk marker for suicide attempts independent of demographic characteristics and co-occurring diagnoses [
19]. These findings suggest that ASC explains additional variance in suicidality, not accounted for by other well-known risk markers in the general population which are more prevalent in ASC, such as depression [
20‐
22] or social isolation [
23,
24], which have been associated with increased risk of suicidality in ASC [
2,
15,
17,
25,
26]. Hence, there may be as yet unknown unique risk markers for suicidality in ASC that are not shared with the general population or other clinical groups, requiring adapted suicide prevention strategies [
6].
Studies exploring the characteristics of suicidality in ASC could provide important clues for possible unique risk markers in this group. For example, the highest rates of suicidal ideation (66%) were reported in adults newly diagnosed with AS, who had struggled without support [
2]. Age of diagnosis and adequate access to post-diagnostic support could therefore be particularly important in preventing suicidality in ASC [
2]. However, many children and adults diagnosed with ASC not only struggle to obtain their diagnosis, but also struggle to obtain post-diagnostic support [
27‐
29]. Lack of tangible social support has been associated with increased risk of suicidality, indirectly through depression [
25].
In the general population, the global male to female ratio of deaths by suicide is estimated to be 1.7 [
30], indicating that males are more likely to die by suicide than females. However, in the one available study exploring death by suicide in the autistic community, autistic females without intellectual disability (ID) were more at risk of dying by suicide (0.32%) compared to autistic males (0.3%); opposite to the general population where males (0.05%) were more likely to die by suicide than females (0.03%) [
15]. Autistic females have been under-researched, and it has been recognised that this group may also be under-diagnosed [
29,
31,
32]. Autistic people report attempting to camouflage their ASC in order to try and fit in in social situations, which may delay obtaining a timely ASC diagnosis and negatively affect their mental health [
31‐
33]. However, no study has quantitatively measured associations between ‘camouflaging’ and risk of mental health difficulties or suicidality in both autistic males and females.
In addition to lack of research into possible autism specific risk markers for suicidality, some potentially common risk factors for suicidality in those with and without ASC diagnosis have very different conceptualisations that have resulted in them being overlooked by researchers and clinicians. For example, self-injurious behaviour in ASC [
34] is conceptualised rather differently than NSSI in the general population, as primarily a restricted and repetitive behaviour characteristic of ASC [
35]. By contrast NSSI in the general population is considered a possible risk marker for later suicide attempts [
9]. Only one study has explored NSSI in autistic adults without co-occurring ID using a tool validated for online research in non-clinical populations [
10] (non-suicidal self-injury assessment tool (NSSI-AT)) [
36]. The rate of NSSI in ASC was elevated (50%) compared to college students (17%) and adult community samples (23%), but the phenomenology of NSSI was broadly similar between those with and without ASC [
10]. Importantly, this suggests that NSSI could be more prevalent in ASC than that in the general population, and could potentially be a previously unexplored common risk factor for suicidality in ASC and the general population.
Previous research has taken a piecemeal approach to furthering our understanding of suicidality in ASC. Important limitations include the fact that no suicidality studies in ASC have used a suicidality assessment tool with evidence of validity in this group [
7,
8], and very few studies have included a comparison group [
3]. Studies have also failed to disentangle common shared and unique risk markers for suicidality in autistic and general populations, which is key to understanding and preventing suicide in ASC [
6].
The current study thus aimed to address these pitfalls in previous suicidality in ASC research. First, we used both a review of the available literature, and consultation with a steering group of autistic adults who have experienced suicidality, to ensure that we identified a range of high priority risk markers for suicidality in autism, some of which may be unique to this group. Second, we are the first to utilise a well-validated suicidality assessment tool (the Suicide Behaviours Questionnaire-Revised (SBQ-R)) [
37] in autistic adults (confirmed in a systematic review) [
7], and NSSI assessment tool previously utilised in autistic adults (NSSI-AT) [
10]. We also include a general population comparison group. Hence, we are able to explore whether autistic adults are at increased risk of suicidality compared to the general population, while controlling for known common risk factors for suicidality (e.g. age, sex, mental health problems, employment, living situation). We also explore for the first time a potentially unique risk marker for suicidality and NSSI in ASC males and ASC females—camouflaging ASC in order to cope in social situations—as well as age of ASC diagnosis, and unmet support needs. We also explore whether NSSI is an independent risk marker for suicidality in those with and without ASC, and whether autistic traits are an independent risk marker for suicidality in the general population without ASC diagnosis.
Discussion
Previous research exploring suicidality in ASC has failed to include adequately sized samples, matched comparison groups, explore risk or protective factors [
2,
3,
6], or include validated suicidality assessment tools [
7]. The current study aimed to address these weaknesses of previous research, to identify common and unique risk markers for suicidality in ASC. Specifically, whether there are unique aspects of ASC and autistic traits that increase risk of suicidality, after statistically controlling for common risk factors such as age, sex, employment, or mental health. We then explored possible unique risk factors which could explain increased risk of suicide in ASC, identified by our steering group of autistic adults: camouflaging one’s ASC in an attempt to fit in in social situations, age of ASC diagnosis, whether people felt they received the support they required, and NSSI. Previous studies have not systematically studied unique and common risk markers for suicidality in ASC compared to the general population, which has prevented development of tailored suicide prevention strategies for this group [
6].
Results are consistent with previous findings that autistic adults are at significantly increased risk of suicidality compared to the general population [
2]. A majority (72%) of autistic adults scored significantly above the recommended cut-off for suicide risk in psychiatric populations, significantly higher than general population adults (33%) with similar age and gender composition. This significant association between ASC diagnosis and suicidality remained when controlling for a number of demographics and diagnoses, known to increase or decrease risk of suicidality in the general population (employment, depression, anxiety, and satisfaction with living arrangements). Additionally, the significant association between self-reported autistic traits in the general population and suicidality remained after statistically controlling for these demographics and diagnoses. These results suggest that autism diagnosis and autistic traits explain significant additional variance in suicidality beyond a range of known risk factors, and are therefore independent risk markers for suicidality. This is consistent with research showing that ASC diagnosis is an independent risk marker for suicide attempts when controlling for a range of demographics and co-occurring diagnoses [
19]. These findings suggest additional unique contributors to suicidality in ASC, which must be addressed in addition to important well-known factors such as mental health, employment, and living arrangements.
The current study explored a potentially unique risk marker for suicidality in ASC, identified by our steering group of autistic adults: tendency to camouflage one’s ASC in order to cope in social situations. Previous research [
29,
32] and discussions with our steering group identified camouflaging as an important potential barrier to timely ASC diagnosis, and having a negative impact on mental health and risk of suicidality. Previous research has also suggested that camouflaging is primarily experienced by autistic females [
31,
33], which may at least in part explain why this group has been under-diagnosed [
43]. Results from the current study however showed subtle differences in camouflaging behaviour between autistic males and females: there was no sex difference in reporting whether one engages in camouflaging behaviour, but autistic females tended to report that they camouflaged across more situations, more frequently and more of the time than autistic males.
Camouflaging significantly predicted suicidality in the ASC group, after controlling for age, sex, presence of at least one developmental condition, depression, anxiety, employment, and satisfaction with living arrangements. Camouflaging and age of ASC diagnosis, and suicidality and age of ASC diagnosis were not significantly correlated. This suggests that camouflaging is directly associated with suicidality rather than in combination with delay in ASC diagnosis. Camouflaging also explained significant additional variance in suicidality above depression or anxiety, suggesting that the association with suicidality is, at least in part, independent of mental health. This is the first evidence of camouflaging being a unique independent risk factor for suicidality in ASC.
In order to engage in camouflaging, one must have insight into one’s own difficulties, how these may be negatively perceived by others, and have a strong motivation to adapt one’s social behaviour to be accepted. Understanding associations between these factors with camouflaging, and the consequent impact on mental health would be valuable. For example, autistic people who have greater insight into their own difficulties are more likely to be depressed than those with less insight [
44], and autistic people are able to accurately predict how family members perceive them, despite being different to their own view [
45]. It would be interesting to explore whether perspective taking ability and insight into one’s own difficulties increase likelihood of engaging in camouflaging behaviour with consequent negative impact on mental health and suicidality.
Importantly, our findings challenge the assumption that autistic people are socially unmotivated, consistent with calls for more accurate and useful autism research, embracing the unique nature of social interest in autism [
46]. It is perhaps more accurate to acknowledge a “double empathy problem”, where autistic people are misinterpreted by non-autistic people and vice versa [
45,
47,
48], which contribute to feelings of isolation among autistic people [
49]. Increasing acceptance of autistic people in society could therefore lead to a reduced need for camouflaging and increased feelings of belonging—a protective factor for suicidality [
17,
23].
Contrary to expectations, and discussions with our autistic steering group, age of ASC diagnosis was not significantly correlated with any other variables, such as mental health problems, suicidality, or NSSI. However, this may have been due to the fact that the mean age of ASC diagnosis was 34 years, and therefore, participants represent autistic people diagnosed in adulthood. Future research will need to explore whether those diagnosed in childhood are significantly less likely to experience mental health problems of suicidality compared to those diagnosed in adulthood. Another important theme identified from discussions with our steering group was lack of access to support, which could compound mental health difficulties and suicidality. Previous research has shown that the autistic community is disconnected from psychiatric services [
18], as many practitioners are not trained in ASC [
50]. The current study therefore quantitatively explored the mismatch between the number of areas an individual would ideally like support, compared to the number of areas they actually received support. These unmet support needs significantly predicted suicidality in the ASC group when controlling for the aforementioned variables. Hence, a clear recommendation for policy and practice to reduce suicide risk in autistic adults, a high-risk group for dying by suicide [
15], is to urgently identify and address unmet support needs in this group. Meeting this shortfall in support could, at least in part, help reduce high rates of suicidality and death by suicide in the autistic community. Research from our group is exploring in more depth barriers and enablers in accessing treatment and support in autistic adults, to help assist in service planning.
The rate of NSSI in the ASC group (63.6%) was significantly higher than the general population group (29.8%), and similar to the rate reported in previous research [
10] (50%), which also utilised the NSSI-AT in autistic adults. NSSI also significantly predicted suicidality in autistic adults, after controlling for a range of known risk factors. Hence, NSSI should not continue to be overlooked, or seen as part of ASC, and rather must be addressed in its own right. Our findings are therefore an important call to action for the research community and clinicians to increase understanding and support for those with ASC experiencing NSSI. However, future studies will need to explore whether this rate of NSSI in ASC adults remains stable, and explore the measurement properties of NSSI assessment tools in ASC.
The current study has a number of strengths as well as limitations. This study is the first to use measures of suicidality (SBQ-R) and NSSI (NSSI-AT) that have good evidence of validity, albeit not yet in autistic adults [
7,
10]. There is a paucity of validated outcome measures for autistic adults, and using tools validated for the general population is an important stop gap until tools adapted for autistic people become available [
7,
10,
51‐
53]. The current study was only cross-sectional, and it is unclear for example whether unmet support needs are a cause or consequence of suicidality. The current study focused on adults, without intellectual disability (ID), and it is unknown whether autism and autistic traits would similarly be a unique risk marker for those with co-occurring ID. Although autism, autistic traits, unmet support needs, and camouflaging explained significant additional variance in suicidality when statistically controlling for a number of other factors, the additional variance explained was small.
ASC diagnosis was assessed by self-report only; however, a majority of participants confirmed the clinic where this diagnosis was obtained. Lifetime suicide attempts in the general population (8%) and ASC group (38%) are similar to previous studies [
2,
17], which suggests that the sample was not biased in this respect. However, lifetime experience of depression in the general population (44.9%) and ASC group (80%) were much higher than previous estimates [
2,
22,
54], despite participants not being recruited because of experience with mental health problems. The rate of mental health difficulties in the current sample therefore may not be representative of the general or autistic populations. A majority of participants in the steering group and online survey were female. Therefore, it could be argued that the topics explored in the survey and study findings apply mostly to autistic females and may not be generalisable to autistic males. However, a majority of autistic males and autistic females reported camouflaging, and regression analyses statistically controlled for sex, suggesting this and other risk markers apply to both sexes.
A key strength and novel aspect of the current study was the participatory research element with a group of autistic adults, who refined the focus of the study, and the content of the survey. This ensured that the study included a range of possible unique and common risk factors for suicidality not explored or considered in previous research on this topic. It also ensured high content validity of the survey, which was refined through three iterations of feedback from the steering group. Previous research has shown that the views of the autistic community which the research affects are rarely included, which can hamper the potential benefits of ASC research for the wider community [
11]. Our study demonstrates the importance of including the voices of autistic people in important and sensitive research that can impact their lives.